PC 4GB RAM limit

In message David Maynard
wrote:

DevilsPGD wrote:
In message David Maynard
wrote:


(Windows x64

I don't know what capabilities and bugs might exist in a pre beta O.S.


You must not get much news up whereever it is you live, Windows x64 has
been in beta for over a year and went RTM several weeks ago.

In short, it hasn't been pre-beta for many many months.


Right.

The problem was me doing a half dozen things at the same time and thinking
of something other than XP/64.

Fair enough, sorry if I was an ass about it


--
It's always darkest before dawn. So if you're going to
steal your neighbor's newspaper, that's the time to do it.

On Fri, 13 May 2005 10:13:25 +0100, Tim Anderson wrote:

The cheaper motherboards (ie. most of them) for P4 or Athlon 64 support a
maximum of 4GB RAM, even if the motherboards and processors support the
AMD64 extensions that can address sqillions of GB. On the other hand, more
upmarket boards support more RAM - up to 24GB or maybe more - for Xeon and
Opteron.

If you buy one of these 4GB boards and install 4GB RAM, you don't get the
use of all of it. The top of the 4GB address space gets shadowed by system
functions such as PCI Express addressing. This is not just a small detail
- typically you lose 1GB of your 4GB. See:

http://www.itwriting.com/blog/?postid=152

I can't at the moment find a clear explanation of this. I understand about
the shadowing, but the question of course is why a modern board can't use
a higher range of addresses to make the full 4GB available to the OS. The
manufacturers mutter about "PC Architecture", but then again they also
make boards that *do* overcome this limit.

With PAE, PC processors have been able to address more than 4GB for years.
So why are we still running into this limit?

Tim

In an Athlon 64 system the limitation is a physical limitation of the
DIMMs not a limitation of the motherboard. Opterons use registered DIMMS.
The register increases latency but it reduces the load on the CPU's
address lines. Because of the reduced loading it's possible to put more
RAMs on a registered DIMM then on an unregistered DIMM. You can buy 2G
registered DIMMs today which means that you can put 8G on an Opteron. The
biggest on buffered DIMM available today is only 1G which is why there is
a 4G limit for the Athlon 64. If/When 2Gbit DDR RAMs become available the
maximum memory size on an A64 system will increase to 8G.

Tim Anderson wrote:
"David Maynard" wrote in message
...


I see. Well, since you hadn't mentioned it again I threaded back to see if
you said what board it was and from the docs it appears to not be a '64
bit' board. By that I mean, it apparently doesn't have anything to support
64 bit specific features and, in particular, no more than the traditional
4 GB address space, which could explain why it doesn't remap HiMem.


Not entirely true. The board specifically supports processors with EM64T
extensions.

I know it supports 'the processor', and it is not unusual to add 'support'
for new processors, but that doesn't mean it was originally intended to
support more than 4 gig.


The literature targets it at 'desktops' and, specifically, XP/2000 as the
supported O.S. and neither of them would be able to use remapped memory
even if the board did it, which might also explain why they didn't bother.


OTOH Intel offers downloads *for this board* for WinXP x64. So apparently
Intel does expect this to be an OS that you might use with this board.

There is no reference to '64 bit' in any of the literature (other than
obliquely via listing processors that work with it) and that they now offer
downloads to allow XP/64 doesn't mean the board was intended nor *designed*
to be a 64 bit platform.

So this isn't the explanation.

I disagree.


Tim

DevilsPGD wrote:

In message David Maynard
wrote:


DevilsPGD wrote:

In message David Maynard
wrote:



(Windows x64

I don't know what capabilities and bugs might exist in a pre beta O.S.


You must not get much news up whereever it is you live, Windows x64 has
been in beta for over a year and went RTM several weeks ago.

In short, it hasn't been pre-beta for many many months.


Right.

The problem was me doing a half dozen things at the same time and thinking
of something other than XP/64.


Fair enough, sorry if I was an ass about it

No problem. My head was on the rinse/spin cycle at the time

"David Maynard" wrote in message
...

I know it supports 'the processor', and it is not unusual to add 'support'
for new processors, but that doesn't mean it was originally intended to
support more than 4 gig.

It is not advertised as "more than 4 gig". It is advertised as 4 gig; but in
fact only offers 3 usable GB.

There is no reference to '64 bit' in any of the literature (other than
obliquely via listing processors that work with it) and that they now
offer downloads to allow XP/64 doesn't mean the board was intended nor
*designed* to be a 64 bit platform.

I don't follow you here. Intel advertise the board as EM64T-compatible. It
supports x64 operating systems for this board with downloadable drivers. Why
would it do that, if it didn't intend you to use a 64-bit OS?

Even Intel support didn't offer this to me an an explanation, when I called.

So this isn't the explanation.

I disagree.

Fair enough, though puzzling in the face of the evidence.

Tim

"General Schvantzkoph" wrote in message
news
On Fri, 13 May 2005 10:13:25 +0100, Tim Anderson wrote:

In an Athlon 64 system the limitation is a physical limitation of the
DIMMs not a limitation of the motherboard.

Not altogther. Athlon systems have the same problem with the address space
between 3GB and 4GB. However, some of the boards offer a remapping function.
Eg. this function in the BIOS of the MSI K8N SLI Platinum:

S/W & H/W memory hole Remapping
This field enables software/hardware to remap the physical memory to the
address higher than 00E0. (This item only activities in 64-bit OS)
Setting options:
[Disabled], [Enabled].

Tim

On Tue, 17 May 2005 00:10:04 +0100, "Tim Anderson"
wrote:


"David Maynard" wrote in message
...

I know it supports 'the processor', and it is not unusual to add 'support'
for new processors, but that doesn't mean it was originally intended to
support more than 4 gig.

It is not advertised as "more than 4 gig". It is advertised as 4 gig; but in
fact only offers 3 usable GB.


While I agree that this disparty is unacceptible, that's
also how it's been for years now. Boards adverstised as
PC3200 (or PC133, whatever...) and 1.5GB, etc, support but
when it actually comes time to install the modules, that's
not how it works out. In the best scenarios one could just
manually set memory timings slower than spec'd (or the
board's bios did it automatically, as was very common on OEM
boxes). Granted that was a stability issue rather than
logical, but end result was same- one couldn't expect to use
the absolute max value memory stated.

Tim Anderson wrote:
"David Maynard" wrote in message
...


I know it supports 'the processor', and it is not unusual to add 'support'
for new processors, but that doesn't mean it was originally intended to
support more than 4 gig.


It is not advertised as "more than 4 gig". It is advertised as 4 gig; but in
fact only offers 3 usable GB.

And it *does* have 4 Gig addressing capability.

That the system area occupies upper memory is characteristic of the
standard PC architecture and will be a fact in any system with a 4 gig
total address space.


There is no reference to '64 bit' in any of the literature (other than
obliquely via listing processors that work with it) and that they now
offer downloads to allow XP/64 doesn't mean the board was intended nor
*designed* to be a 64 bit platform.


I don't follow you here. Intel advertise the board as EM64T-compatible.

It's your interpretation of what 'compatible' means that causes the confusion.

EM64T processors operate in the board, within the 4 GB address space
limitation. It is, indeed, 'compatible' with them.


It
supports x64 operating systems for this board with downloadable drivers. Why
would it do that, if it didn't intend you to use a 64-bit OS?

You confuse the 'intent' of a designer, when the board is designed, vs the
'intent' of making new things available within the context, and
limitations, of an already existing design.

For example, I have (32 bit) XP, a 4GB (including system area) capable
operating system running on motherboards that support no more than 768 Meg
of RAM but you don't see me claiming the board 'should' handle 4 GB simply
because they made XP drivers available for it so how come you think a
motherboard designed with a 4 GB address space should magically find a way
to map 1 GB into the un-addressable area above the 4 GB address space just
because Intel made drivers available?

Would you be happier if they just didn't provide support for newer
processors and operating systems?

Even Intel support didn't offer this to me an an explanation, when I called.

So this isn't the explanation.

I disagree.


Fair enough, though puzzling in the face of the evidence.

It isn't puzzling to a design engineer.

As for "the evidence," I look at sales brochures, and the like, to see at
what target market it appears to be aimed. For example, this sales brochure
of the entire line that motherboard comes from:

http://cache-www.intel.com/cd/00/00/...752_147752.pdf

There is no mention of EM64T or '64 bit computing' anywhere in it. Seems
odd they would not mention this in the "Features and Benefits" if they had
(originally) 'intended' this to be a 'feature and benefit'.

Also, the 'boxed' page for that board (reseller's page with what goes into
web sites and such for their 'sales literature') doesn't mention EM64T
either (unless you deduce it through the supported processors links).

http://www.intel.com/cd/channel/rese...5pbl/index.htm

When you get to the technical documentation
(http://www.intel.com/design/motherbd/bl/index.htm)
EM64T 'support' becomes more obvious, but then a system builder should also
read all of it.


Tim

"David Maynard" wrote in message
...
Tim Anderson wrote:
"David Maynard" wrote in message
...

It is not advertised as "more than 4 gig". It is advertised as 4 gig; but
in fact only offers 3 usable GB.

And it *does* have 4 Gig addressing capability.

I think I have a good understanding of this issue now. There is an inherent
problem with this top 1GB of address space. It is possible for boards to
overcome it by remapping. This board doesn't though. Personally, I think it
should; and I think Intel should make the problem clearer; but it's no big
deal. I raised it here because I wanted to understand it better.

When you get to the technical documentation
(http://www.intel.com/design/motherbd/bl/index.htm)
EM64T 'support' becomes more obvious, but then a system builder should
also read all of it.

It either supports it or it doesn't, no matter what the marketers say or
don't say. I'd have thought a design engineer would understand that :-)

Tim

Tim Anderson wrote:
"David Maynard" wrote in message
...

Tim Anderson wrote:

"David Maynard" wrote in message
...


It is not advertised as "more than 4 gig". It is advertised as 4 gig; but
in fact only offers 3 usable GB.

And it *does* have 4 Gig addressing capability.


I think I have a good understanding of this issue now. There is an inherent
problem with this top 1GB of address space. It is possible for boards to
overcome it by remapping. This board doesn't though. Personally, I think it
should;

It can't because a remap would place it outside of 4 gig and it can only
address 4 gig.

Remember, we're talking *physical* address space and to physically remap
memory above 4 gig the board would need at least one more address line,
which would mean it could address 8 gig. But it isn't an 8 gig motherboard.

(I'm not sure what's actually 'missing' on the board because the chipset
specs suggest the chipset itself could address 8 gig but 'something' is
apparently missing. The point being adding that 'feature' may not be as
trivial as it seems.)

and I think Intel should make the problem clearer;

I agree with that. The only explanations I can think of off the top of my
head are 1. they didn't really expect folks putting 4 gig in it would be
all that common and/or 2. their primary market is system
builders/manufacturers who are expected to read the detailed documentation.

but it's no big
deal. I raised it here because I wanted to understand it better.


When you get to the technical documentation
(http://www.intel.com/design/motherbd/bl/index.htm)
EM64T 'support' becomes more obvious, but then a system builder should
also read all of it.


It either supports it or it doesn't,

That depends on what "it" is.

If "it" is being able to run an EM64T processor in 64 bit mode then the
board supports "it."

If "it" is having an address space larger than 4 gig then the board does
not support "it."

Now, the second 'it', being what? a terabyte?, is not going to be 'fully'
implemented no matter what so you *know* there has to be a limit
*somewhere* with *any* board which 'supports' EM64T. It happens to be 4 gig
on this one.


no matter what the marketers say or
don't say.

Seems to me you might want to consider this example a warning that you need
to pay more attention to what they say

I'd have thought a design engineer would understand that :-)

A design engineer understands what he's told to design and, when designing
a motherboard in the 2003-2004 time period, putting in 'support' for things
that don't yet exist isn't necessarily a part of it. And if the spec says
"design a P4 motherboard with 4 gig address space" then that's what he'll
design, if he's smart.

And, btw, I'm not being frivolous about that. One of the biggest problems
design engineers have is folks coming in after it's all said and done
complaining "how come you didn't include X?"

Because it WASn't IN the SPEC!

The second biggest problem is the poor engineer who thought he'd be clever
and include some neat things having to explain why he's wasting resources
on something that WASn't IN the SPEC.

And I tell ya, it's usually a heck of a lot easier to answer number 1 than
it is to explain number 2.


Tim